We are planning to do load testing of our file feeds on our Integration Server 4.6. We use both flow services & java in general. What are the most important points that we should keep in mind when developing efficient services(flow/java services) in the integration server? One of the important practice is to create as few strings as possible, because they are immutable & memory intensive. It would greatly help if you could highlight other aspects that we should take care of, especially when trying to process files(data transformation) with more than 30000 records in a flat file into a database(data loading). I'm talking about batch processing as apposed to online transactions.